A word segmentation model for multi-department early warning information and its quality control method

Abstract

This invention provides a method, apparatus, device, and medium for quality control of multi-departmental early warning information content. The method includes: receiving early warning information messages, parsing and obtaining a unique department index; matching an appropriate early warning word segmentation model container service for the corresponding department using the department index; extracting the early warning content and title, and matching the applicable release and handling stage according to the rule attributes of the quality control rule set in the early warning word segmentation model container service of the corresponding department; executing all quality control rules in the quality control rule set to obtain the processing result of each quality control rule; determining whether there are any non-compliance identifiers in the quality control result set; if so, blocking the release and integrating and outputting the non-compliance with the quality control rule description; otherwise, proceeding to the matched release and handling stage. This invention can provide unified content control for various early warning information from multiple departments without requiring each department to develop its own control system for early warning information content.

Description

Technical Field

[0001] This invention relates to the field of analysis technology for the control of early warning information content, and in particular to a quality control model and method for early warning information content from multiple departments. Background Technology

[0002] Early warning information is a medium through which responsible units provide free information dissemination and defense guidance to the public. It includes attributes such as the issuing unit, the content of the warning information, the timeliness of the warning, the warning level, the warning title, and the release time. It covers various types of disasters, including natural disasters, accidents, public safety, social security, and risk information. Effectively improving the timeliness and accuracy of early warning information dissemination ensures that information is delivered to emergency response personnel and government decision-makers as quickly as possible to mitigate risks and guide disaster prevention and mitigation efforts. Therefore, the accuracy and authority of early warning information content are crucial.

[0003] Currently, early warning information agencies at all levels act as bridges for information exchange between various departments, aggregating early warning information from multiple departments and disseminating it precisely to risk-covered areas through a unified network of channels. The access to early warning information from various departments is highly personalized, with complex business rules. Currently, both manual and system-based review methods are used to ensure the quality of early warning information dissemination. Manual review requires staff to be highly proficient in the early warning procedures of each department, but instances of inaccurate review, inconsistencies in information content, and incorrect wording still occur. At the system application level, judging business quality control and time quality control in early warning attribute fields through static pages is not conducive to the flexible expansion of early warning quality control rules and the adaptation to the business rules of multiple departments. Furthermore, the lack of digitized methods for handling issues such as business quality control correlations, reduplicated words, and wording mismatches in early warnings from multiple departments, coupled with the impact of manual review on information processing efficiency, further hinders this process.

[0004] Currently, most early warning systems lack content verification for warnings, relying primarily on the professional competence and rigorous review of the personnel issuing the warnings, or on rules agreed upon during program development to verify the content. This approach has the following drawbacks:

[0005] 1. The warning keyword database relies on manual maintenance, resulting in slow updates;

[0006] 2. Traditional rules may not be applicable to every department, leading to situations that affect their use;

[0007] 3. Lack of effective rule derivation means that new rules cannot meet the actual situation;

[0008] 4. Traditional rules lack integration with a rule engine, and adding or modifying rules will affect the real-time use of the business system.

[0009] Chinese invention patent CN202211688001.0 discloses a method and system for reviewing meteorological warning information based on natural language processing. The method involves acquiring the meteorological warning text to be reviewed, inputting it into a trained text review model to obtain semantic classification results; segmenting the text into words to obtain a segmentation sequence; parsing the segmentation sequence based on a configured set of syntactic rules to obtain parsing results; the syntactic rule set includes syntactic rules corresponding to different warning sentence structures; and determining the review result of the meteorological warning text based on the parsing and classification results. This method overcomes the problems of long processing times and susceptibility to errors in manual review, achieving accurate and comprehensive review of meteorological warning text. However, this review method and system only target meteorological warning information from a single meteorological department and cannot handle the unified control of various warning information contents from multiple departments. Therefore, for different departments (such as meteorological departments, water resources departments, earthquake departments, etc.), each department must develop its own control system suitable for its own warning information content. Summary of the Invention

[0010] The technical problem to be solved by the present invention is to provide a quality control model and method for the content of early warning information from multiple departments, which can perform unified control of the content of various early warning information from multiple departments, without the need for each department to develop a control system suitable for its own early warning information content.

[0011] In a first aspect, the present invention provides a quality control method for multi-departmental early warning information content, comprising the following steps:

[0012] S1. Receive the message of the early warning information, intercept and parse the structure of the message and obtain the unique department index; match the early warning word segmentation model container service that is suitable for the corresponding department through the department index;

[0013] S2. Extract the warning content and warning title from the structure, and match the warning content and warning title to the release and handling stage applicable to the quality control rules according to the rule attributes of the quality control rule set in the warning word segmentation model container service of the corresponding department;

[0014] S3. Execute all quality control rules in the quality control rule set on the warning content and the warning title to obtain the processing result of each quality control rule. Add each processing result to the quality control result set after matching it with an identifier. The quality control rules in the quality control rule set include time verification rules and content verification rules. The identifier is pass, fail, or friendly reminder.

[0015] S4. Determine whether there are any non-passing identifiers among all the processing results in the quality control result set. If so, prevent the release program from continuing to execute and integrate and output the descriptions of non-compliance with the quality control rules. If not, proceed to the matching release processing stage.

[0016] Secondly, the present invention provides a quality control device for multi-department early warning information content, comprising:

[0017] The department adaptation module is used to receive the early warning information message, intercept and parse the structure of the message and obtain a unique department index; and match the early warning word segmentation model container service corresponding to the department through the department index.

[0018] The release and handling adaptation module is used to extract the warning content and warning title from the structure; and to match the warning content and warning title to the release and handling stage applicable to the quality control rules according to the rule attributes of the quality control rule set in the warning word segmentation model container service of the corresponding department.

[0019] The content quality control module is used to execute all quality control rules in the quality control rule set on the warning content and the warning title, obtain the processing result of each quality control rule, and add each processing result to the quality control result set after matching it with an identifier; the quality control rules in the quality control rule set include time verification rules and content verification rules; the identifier is pass, fail, or friendly reminder;

[0020] The quality control result processing module is used to determine whether there are any non-compliance indicators among all the processing results in the quality control result set. If so, the release program is prevented from continuing to execute, and the descriptions of non-compliance with the quality control rules are integrated and output. If not, the matching release processing stage is entered.

[0021] Thirdly, the present invention provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method described in the first aspect.

[0022] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in the first aspect.

[0023] The one or more technical solutions provided in this embodiment of the invention have at least the following technical effects or advantages: Through the pre-set warning word segmentation model container service for each department, warning information from different departments (such as meteorological departments, earthquake departments, fisheries departments, epidemic control departments, etc.) can be uniformly received and processed with unified quality control. Furthermore, during the word segmentation stage, a multi-department word segmentation model is used, with each department's model performing word segmentation based on the corresponding department's word usage characteristics. This forms a personalized word segmentation container for the output of multi-department warning information and risk messages, improving word segmentation efficiency while further enhancing the content quality of multi-department warning information releases.

[0024] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0025] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0026] Figure 1 This is a schematic diagram of the system framework of the present invention;

[0027] Figure 2 This is a flowchart of the method in Embodiment 1 of the present invention;

[0028] Figure 3 This is a schematic diagram of the quality control rule set in an embodiment of the present invention;

[0029] Figure 4 This is a flowchart of the word segmentation operation for the warning content in an embodiment of the present invention;

[0030] Figure 5 This is an example schematic diagram of a directed acyclic graph in an embodiment of the present invention;

[0031] Figure 6 This is a schematic diagram of the device in Embodiment 2 of the present invention;

[0032] Figure 7 This is a schematic diagram of the electronic device in Embodiment 3 of the present invention;

[0033] Figure 8 This is a schematic diagram of the structure of the medium in Embodiment 4 of the present invention. Detailed Implementation

[0034] This application provides a quality control model and method for multi-department early warning information content, which can uniformly control the content of various early warning information from multiple departments without requiring each department to develop its own control system for early warning information content.

[0035] The overall concept of the technical solution in this application is as follows: This application utilizes a pre-defined early warning word segmentation model container service for various departments to uniformly receive early warning information from different departments (such as meteorological departments, earthquake departments, fisheries departments, epidemic control departments, etc.) and perform unified quality control processing. Furthermore, during the word segmentation stage, a multi-department word segmentation model is used. This model performs word segmentation based on the word usage characteristics of the corresponding departments of the early warning information, forming a personalized word segmentation container for the output of multi-department early warning information and risk messages. This improves word segmentation efficiency while further enhancing the content quality of multi-department early warning information releases.

[0036] Before introducing specific embodiments, the system framework corresponding to the method in the embodiments of this application will be introduced first, such as... Figure 1 As shown, the system is roughly divided into several parts: an early warning center and various departmental systems.

[0037] The early warning center is used to receive early warning information from different departments (such as meteorological departments, earthquake departments, fishery departments, epidemic control departments, etc.) and perform unified quality control processing; it has a multi-department early warning word segmentation model and a set of quality control rules.

[0038] The multi-department early warning word segmentation model is trained and segmented based on historical early warning information categorized by department using a basic lexicon. It then employs natural language word segmentation to segment real-time incoming departmental early warning information according to departmental characteristics, resulting in more accurate segmentation. The multi-department early warning word segmentation model constructs initial word segmentation containers for each department to store the word segmentation models adapted to each department. If early warning information from a new department is received, a new word segmentation container for that department is automatically created.

[0039] The quality control rules in the quality control rule set include time verification rules and content verification rules. The content verification rules may specifically include business verification rules, reduplicated word verification rules, symbol verification rules, and NLP word segmentation verification rules.

[0040] Each department or system that needs to issue early warning information, such as meteorological departments, earthquake departments, fishery departments, water conservancy departments, and epidemic prevention and control departments, will send its own early warning information to the early warning center, which will then conduct unified quality control processing before issuing the warning information.

[0041] Example 1

[0042] like Figure 2 As shown, this embodiment provides a quality control method for multi-department early warning information content, including the following steps:

[0043] S1. Receive the message of the early warning information, intercept and parse the structure of the message and obtain the unique department index; match the early warning word segmentation model container service that is suitable for the corresponding department through the department index.

[0044] The structure of the early warning message includes a code that indicates the department to which the early warning message belongs.

[0045] S2. Extract the warning content and warning title from the structure, and match the warning content and warning title to the release and handling stage applicable to the quality control rules according to the rule attributes of the quality control rule set in the warning word segmentation model container service of the corresponding department;

[0046] The release and processing process generally includes multiple steps such as data entry, review, issuance, and verification.

[0047] S3. Execute all quality control rules in the quality control rule set on the warning content and the warning title to obtain the processing result of each quality control rule. Add each processing result to the quality control result set after matching it with an identifier. The quality control rules in the quality control rule set include time verification rules and content verification rules. The content verification rules may specifically include business verification rules, reduplicated word verification rules, symbol verification rules, and NLP word segmentation verification rules. Specific rule names are as follows: Figure 3 As shown; the identifier indicates pass, fail, or friendly reminder;

[0048] The time-based rules extract time data from the text content and make judgments based on the current time or the publication time in the information body, thereby performing quality control from the time dimension. The text content quality control rules, such as business verification rules, reduplicated word verification rules, symbol verification rules, and NLP word segmentation verification rules, will first process the content through a multi-department word segmentation model before verification. If the text content quality control rules are not met, the description of the failure to meet the quality control rules will be recorded in the result set object, and the next quality control rule will be executed regardless of whether the rules are met or not.

[0049] S4. Determine whether there are any non-passing identifiers among all the processing results in the quality control result set. If so, prevent the release program from continuing to execute and integrate and output the descriptions of non-compliance with the quality control rules. If not, proceed to the matching release processing stage.

[0050] Among them, such as Figure 4 As shown, as a more preferred or more specific implementation of this embodiment, before performing verification through the content verification rules, the warning content must first be segmented into words. The word segmentation operation includes the following process.

[0051] S11. Access any departmental early warning information, parse the content of the departmental early warning information according to the standard parsing protocol, perform preliminary segmentation into different sentence segments through regular expressions and string matching, and automatically call the corresponding department's word segmentation container for storage;

[0052] The word segmentation container is constructed by loading a basic lexicon file and obtaining word groups related to the characteristics of the corresponding department from the basic lexicon.

[0053] The initial segmentation is achieved by using a predefined regular expression and relevant flags to initially divide long segments of text. It checks whether a string contains a certain substring. If so, the matching substring is replaced or a substring that meets a certain condition is extracted from a string.

[0054] S12. Construct all possible word segmentation scenarios for each sentence segment and generate a personalized directed acyclic graph containing multi-departmental early warning information, thus generating a directed acyclic dataset for each sentence segment; specifically:

[0055] The content of the departmental early warning information is parsed according to the standard parsing protocol. It is initially segmented into different sentence segments using regular expressions and string matching. Storage variables are defined for each sentence segment set. By traversing and changing the position of each character in the sentence segment, character combinations are obtained. At the same time, the character combinations are compared with the word groups in the corresponding department's word segmentation container to determine whether the character combinations are common word groups. This process is repeated until all character combinations in the sentence segment have been judged. Finally, a related directed acyclic graph is generated.

[0056] For example, a sentence might be: "Fujian Provincial Department of Water Resources issued a statement." This sentence is segmented into individual words, and each word is combined to generate corresponding word groups. Finally, a directed acyclic dataset is generated based on the position of each word and its combined words, which can be represented as Ex={0:[1,2,3],1:[2],2:[3,5,6],3:[4,5,6],4:[5],5:[6],6:[7,8],7:[8],8:[8]}. Its directed acyclic graph is shown below. Figure 5 As shown.

[0057] S13. Perform maximum probability word segmentation combination calculation on the directed acyclic data set of each sentence segment, and output the maximum probability word segmentation group set for each sentence segment;

[0058] The method for calculating the phrase with the highest probability of each character in a sentence segment is as follows:

[0059]

[0060] Where E x (i) represents the phrase in which the i-th character appears in the sentence segment, P(E x (i) represents the probability of the word group appearing, i.e.: the probability of each word group = the number of word groups of the word in the corresponding department segmentation container / the total number of words in the corresponding department segmentation container; S is the number of characters in the sentence segment, and Pi is the maximum probability of the word group appearing;

[0061] For example, the character "Fu" can be combined with other characters one by one to form multiple basic phrases, including "Fu", "Fujian", and "Fujian Province". According to the above basic phrase-word count key-value pairs, the word frequencies of these three phrases can be obtained for the character "Fu". The probability of "Fu" appearing = the word frequency of "Fu" / the total number of phrases in the department. The probability of the phrase "Fujian" appearing = the word frequency of "Fujian" / the total number of phrases in the department. The probability of the phrase "Fujian Province" appearing = the word frequency of "Fujian Province" / the total number of phrases in the department. Then Pi takes the maximum value of the probabilities of these three phrases appearing.

[0062] S14. After calculating the maximum probability phrases for all single characters in a sentence segment, calculate the maximum probability phrase path R of the sentence segment, that is:

[0063] R = {S: [Pi, E]};

[0064] Where S is the node serial number of the word, Pi is the maximum probability of the phrase combined by the word appearing, and E is the combination serial number of the word, that is, the end position of the maximum probability phrase;

[0065] For example, performing a reverse calculation on the above maximum probability path, the directed acyclic set Ex of "The Water Resources Department of Fujian Province issues" has 8 nodes, as Figure 5 shown. Now perform the maximum probability calculation for each node as follows:

[0066] P(8) = 1;

[0067] P(7) = Max(P(8) * P(7→8));

[0068] P(6) = Max(P(7) * P(6→7), P(9) * P(6→8));

[0069] P(5) = Max(P(6) * P(5→6));

[0070] P(4) = Max(P(5) * P(4→5));

[0071] P(3) = Max(P(4) * P(3→4), P(5) * P(3→5), P(6) * P(3→6));

[0072] P(2) = Max(P(3) * P(2→3), P(5) * P(3→5), P(6) * P(3→6));

[0073] P(1) = Max(P(2) * P(1→2));

[0074] P(0) = Max(P(1) * P(0→1), P(2) * P(0→2), P(3) * P(0→3)); <00001After the above calculations are completed, the maximum probability of each node is obtained. For example, node P(0) has three paths: 0 to 1, 0 to 2, and 0 to 3. The path with the highest probability value is taken as the maximum probability path. Assuming that the maximum probability path for node 0 is R0 = {0, P(0), 2}, it means that the maximum probability value of node 0 is P(0), and the ending position of the maximum probability path is 2. The maximum probability paths for the following nodes are calculated in this way. Finally, the ending position is taken from node 0 to form a phrase. After processing, the system automatically jumps to the beginning of the next word until the end of the sentence.

[0076] After obtaining the maximum probability word path of all segments of the departmental early warning information, all segments of the departmental early warning information are segmented, and then information word segmentation for content quality control of the departmental early warning information is generated.

[0077] Based on the same inventive concept, this application also provides an apparatus corresponding to the method in Embodiment 1, as detailed in Embodiment 2.

[0078] Example 2

[0079] like Figure 6 As shown, this embodiment provides a quality control device for multi-department early warning information content, including:

[0080] The department adaptation module is used to receive the early warning information message, intercept and parse the structure of the message and obtain a unique department index; and match the early warning word segmentation model container service corresponding to the department through the department index.

[0081] The release and handling adaptation module is used to extract the warning content and warning title from the structure; and to match the warning content and warning title to the release and handling stage applicable to the quality control rules according to the rule attributes of the quality control rule set in the warning word segmentation model container service of the corresponding department.

[0082] The content quality control module is used to execute all quality control rules in the quality control rule set on the warning content and the warning title, obtain the processing result of each quality control rule, and add each processing result to the quality control result set after matching it with an identifier. The quality control rules in the quality control rule set include time verification rules and content verification rules. The content verification rules include business verification rules, reduplicated word verification rules, symbol verification rules, and NLP word segmentation verification rules. The specific rule names are as follows: Figure 3 As shown; the identifier indicates pass, fail, or friendly reminder;

[0083] The quality control result processing module is used to determine whether there are any non-compliance indicators among all the processing results in the quality control result set. If so, the release program is prevented from continuing to execute, and the descriptions of non-compliance with the quality control rules are integrated and output. If not, the matching release processing stage is entered.

[0084] Among them, such as Figure 4 As shown, as a more preferred or more specific implementation of this embodiment, the quality control device for multi-department early warning information content in this embodiment further includes a multi-department word segmentation model, which is used to perform word segmentation on the early warning content before verification by the content verification rules. The word segmentation operation includes the following process.

[0085] S11. Access any departmental early warning information, parse the content of the departmental early warning information according to the standard parsing protocol, perform preliminary segmentation into different sentence segments through regular expressions and string matching, and automatically call the corresponding department's word segmentation container for storage;

[0086] The word segmentation container is constructed by loading a basic lexicon file and obtaining word groups related to the characteristics of the corresponding department from the basic lexicon.

[0087] The initial segmentation is achieved by using a predefined regular expression and relevant flags to initially divide long segments of text. It checks whether a string contains a certain substring. If so, the matching substring is replaced or a substring that meets a certain condition is extracted from a string.

[0088] S12. Construct all possible word segmentation scenarios for each sentence segment and generate a personalized directed acyclic graph containing multi-departmental early warning information, thus generating a directed acyclic dataset for each sentence segment; specifically:

[0089] The content of the departmental early warning information is parsed according to the standard parsing protocol. It is initially segmented into different sentence segments using regular expressions and string matching. Storage variables are defined for each sentence segment set. By traversing and changing the position of each character in the sentence segment, character combinations are obtained. At the same time, the character combinations are compared with the word groups in the corresponding department's word segmentation container to determine whether the character combinations are common word groups. This process is repeated until all character combinations in the sentence segment have been judged. Finally, a related directed acyclic graph is generated.

[0090] For example, a sentence might be: "Fujian Provincial Department of Water Resources issued a statement." This sentence is segmented into individual words, and each word is combined to generate corresponding word groups. Finally, a directed acyclic dataset is generated based on the position of each word and its combined words, which can be represented as Ex={0:[1,2,3],1:[2],2:[3,5,6],3:[4,5,6],4:[5],5:[6],6:[7,8],7:[8],8:[8]}. Its directed acyclic graph is shown below. Figure 5 As shown.

[0091] S13. Perform maximum probability word segmentation combination calculation on the directed acyclic data set of each sentence segment, and output the maximum probability word segmentation group set for each sentence segment;

[0092] For each single character in a sentence segment, the phrase with the highest probability of occurrence is calculated s follows:

[0093]

[0094] Where E x (i) is the phrase in which the i-th character appears in the sentence segment, and P(E x (i)) is the probability of the occurrence of this phrase, that is: the probability of each phrase = the number of phrases of this word in the corresponding department word segmentation container / the total number of words in the corresponding department word segmentation container; S is the number of characters in this sentence segment, and Pi is the highest probability of the occurrence of this phrase;

[0095] For example, the character "福" can be combined with other characters one by one to form multiple basic phrases, including "福", "福建", and "福建省". According to the above basic phrase-word count key-value pairs, the word frequencies of these three phrases can be obtained for the character "福". The probability of the occurrence of "福" = the word frequency of "福" / the total number of phrases in this department, the probability of the occurrence of the phrase "福建" = the word frequency of "福建" / the total number of phrases in this department, and the probability of the occurrence of the phrase "福建省" = the word frequency of "福建省" / the total number of phrases in this department. Then Pi takes the maximum value of the occurrence probabilities of these three phrases.

[0096] S14. After calculating the phrases with the highest probability for all single characters in a sentence segment, calculate the path R of the phrase with the highest probability in this sentence segment, that is:

[0097] R = {S: [Pi, E]};

[0098] Where S is the node serial number of the word, Pi is the highest probability of the occurrence of the phrase combined by this word, and E is the combination serial number of this word, that is, the end position of the phrase with the highest probability;

[0099] For example, performing a reverse calculation on the above path with the highest probability, the directed acyclic set Ex of "福建省水利厅发布" has 8 nodes, as Figure 5 shown. Now, calculate the highest probability for each node as follows:

[0100] P(8) = 1;

[0101] P(7) = Max(P(8) * P(7→8));

[0102] P(6) = Max(P(7) * P(6→7), P(9) * P(6→8));

[0103] P(5) = Max(P(6) * P(5→6));

[0104] P(4) = Max(P(5) * P(4→5));

[0105] P(3)=Max(P(4)*P(3→4),P(5)*P(3→5),P(6)*P(3→6));

[0106] P(2)=Max(P(3)*P(2→3),P(5)*P(3→5),P(6)*P(3→6));

[0107] P(1) = Max(P(2) * P(1→2));

[0108] P(0)=Max(P(1)*P(0→1),P(2)*P(0→2),P(3)*P(0→3));

[0109] After the above calculations are completed, the maximum probability of each node is obtained. For example, node P(0) has three paths: 0 to 1, 0 to 2, and 0 to 3. The path with the highest probability value is taken as the maximum probability path. Assuming that the maximum probability path for node 0 is R0 = {0, P(0), 2}, it means that the maximum probability value of node 0 is P(0), and the ending position of the maximum probability path is 2. The maximum probability paths for the following nodes are calculated in this way. Finally, the ending position is taken from node 0 to form a phrase. After processing, the system automatically jumps to the beginning of the next word until the end of the sentence.

[0110] After obtaining the maximum probability word path of all segments of the departmental early warning information, all segments of the departmental early warning information are segmented, and then information word segmentation for content quality control of the departmental early warning information is generated.

[0111] Since the apparatus described in Embodiment 2 of the present invention is an apparatus used to implement the method of Embodiment 1 of the present invention, those skilled in the art can understand the specific structure and variations of the apparatus based on the method described in Embodiment 1 of the present invention, and therefore will not be described again here. All apparatuses used in the method of Embodiment 1 of the present invention fall within the scope of protection of the present invention.

[0112] Based on the same inventive concept, this application provides an electronic device embodiment corresponding to Embodiment 1, as detailed in Embodiment 3.

[0113] Example 3

[0114] This embodiment provides an electronic device, such as... Figure 7 As shown, it includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it can implement any of the embodiments in Example 1.

[0115] Since the electronic device described in this embodiment is the device used to implement the method in Embodiment 1 of this application, those skilled in the art can understand the specific implementation method and various variations of the electronic device in this embodiment based on the method described in Embodiment 1 of this application. Therefore, how the electronic device implements the method in the embodiment of this application will not be described in detail here. Any device used by those skilled in the art to implement the method in the embodiment of this application falls within the scope of protection of this application.

[0116] Based on the same inventive concept, this application provides a storage medium corresponding to Embodiment 1, as detailed in Embodiment 4.

[0117] Example 4

[0118] This embodiment provides a computer-readable storage medium, such as... Figure 8 As shown, a computer program is stored thereon, which, when executed by a processor, can implement any of the embodiments in Example 1.

[0119] The methods, apparatus, equipment, and media provided in this application have at least the following technical effects or advantages: Through pre-set early warning word segmentation model container services for various departments, early warning information from different departments (such as meteorological departments, earthquake departments, fisheries departments, water conservancy departments, epidemic control departments, etc.) can be uniformly received and processed with unified quality control. Furthermore, during the word segmentation stage, a multi-department word segmentation model is used, with each department's model performing word segmentation based on the corresponding department's word usage characteristics. This forms a personalized word segmentation container for the output of multi-department early warning information and risk messages, improving word segmentation efficiency while further enhancing the content quality of multi-department early warning information releases.

[0120] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or systems, or as computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0121] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0122] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0123] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0124] While specific embodiments of the present invention have been described above, those skilled in the art should understand that the specific embodiments described are merely illustrative and not intended to limit the scope of the invention. Equivalent modifications and variations made by those skilled in the art in accordance with the spirit of the invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A quality control method for multi-departmental early warning information content, characterized in that: Includes the following steps: S1. Receive the message of the early warning information, intercept and parse the structure of the message and obtain the unique department index; match the early warning word segmentation model container service that is suitable for the corresponding department through the department index; The departments mentioned are those units that are required to issue early warning information; The types of early warning information include natural disasters, accidents, public safety, social security, and risk messages; The early warning word segmentation model container service is used to uniformly receive early warning information from different departments and perform unified quality control processing. S2. Extract the warning content and warning title from the structure, and match the warning content and warning title to the release and handling stage applicable to the quality control rules according to the rule attributes of the quality control rule set in the warning word segmentation model container service of the corresponding department; S3. Execute all quality control rules in the quality control rule set on the warning content and the warning title to obtain the processing result of each quality control rule. Add each processing result to the quality control result set after matching it with an identifier. The quality control rules in the quality control rule set include time verification rules and content verification rules. Before performing verification according to the aforementioned content verification rules, the warning content must first be segmented into words. Furthermore, the word segmentation stage is based on a multi-department word segmentation model; the multi-department word segmentation model is obtained by training and segmenting historical early warning information according to departmental categories based on a basic lexicon, and natural language word segmentation is used to segment real-time accessed departmental early warning information based on the word usage characteristics of the corresponding department, forming a word segmentation container with departmental personalization for the output of multi-departmental early warning information and risk messages; The multi-department early warning word segmentation model will construct an initial word segmentation container for each department to store the word segmentation model adapted to each department. If early warning information from a new department is received, the word segmentation container for that new department will be automatically created. The word segmentation container is constructed by loading a basic lexicon file and obtaining word groups related to the characteristics of the corresponding department from the basic lexicon. The time verification rules are used to extract time data from the information text content and make judgments based on the current time or the publication time in the information body, thereby performing quality control from the time dimension. The content verification rules include business verification rules, reduplicated word verification rules, symbol verification rules, and NLP word segmentation verification rules; The identifier can be "pass", "fail", or a friendly reminder. S4. Determine whether there are any non-passing identifiers among all the processing results in the quality control result set. If so, prevent the release program from continuing to execute and integrate and output the descriptions of non-compliance with the quality control rules. If not, proceed to the matching release processing stage.

2. The quality control method for multi-department early warning information content according to claim 1, characterized in that: The word segmentation operation includes the following process; S11. Access any departmental early warning information, parse the content of the departmental early warning information according to the standard parsing protocol, perform preliminary segmentation into different sentence segments through regular expressions and string matching, and automatically call the corresponding department's word segmentation container for storage; S12. Construct all possible word segmentation scenarios for each sentence segment and generate a personalized directed acyclic graph with multi-departmental early warning information content, and generate a directed acyclic data set for each sentence segment. S13. Perform maximum probability word segmentation combination calculation on the directed acyclic data set of each sentence segment, and output the maximum probability word segmentation group set for each sentence segment; The method for calculating the phrase with the highest probability of each character in a sentence segment is as follows: ; in Let i be the phrase whose i-th character appears in the sentence segment. The probability of the phrase appearing is: the probability of each phrase = the number of phrases containing that word in the corresponding department's word segmentation container / the total number of words in the corresponding department's word segmentation container; s is the number of characters in the sentence segment, and Pi is the maximum probability of the phrase appearing. S14. After calculating the maximum probability word group for all characters in a sentence segment, the maximum probability word group path R of that sentence segment is obtained, that is: ; Where S is the node number of the word, Pi is the maximum probability of the word combination appearing, and E is the combination number of the word, i.e. the position of the word combination with the maximum probability. Once the maximum probability word path of all segments of the departmental early warning information is obtained, information segmentation for content quality control of the departmental early warning information is generated.

3. The quality control method for multi-department early warning information content according to claim 2, characterized in that: S12 specifically involves: parsing the content of the departmental warning information according to a standard parsing protocol, initially segmenting it into different sentence segments using regular expressions and string matching, defining storage variables for each sentence segment set, obtaining single-character combinations by traversing and changing the position of each character in the sentence segment, and simultaneously judging the single-character combinations against the word groups in the corresponding department's word segmentation container to determine whether the single-character combinations are common word groups, until all single-character combinations in the sentence segment have been judged, and finally generating a related directed acyclic graph.

4. The quality control method for multi-department early warning information content according to claim 2, characterized in that: In step S11, the word segmentation container is constructed by loading a basic lexicon file and obtaining word groups related to the characteristics of the corresponding department from the basic lexicon. The initial segmentation is achieved by using a predefined regular expression and relevant flags to initially divide long segments of text. It checks whether a string contains a certain substring. If so, the matching substring is replaced or a substring that meets a certain condition is extracted from a string.

5. A quality control device for multi-department early warning information content, characterized in that: include: The department adaptation module is used to receive the early warning information message, intercept and parse the structure of the message, and obtain a unique department index. The system matches the corresponding department with a pre-warning word segmentation model container service using a department index; the department refers to the unit or organization that needs to issue pre-warning information; the types of pre-warning information include natural disasters, accidents, public safety, social security, and risk messages; The early warning word segmentation model container service is used to uniformly receive early warning information from different departments and perform unified quality control processing. The release and handling adaptation module is used to extract the warning content and warning title from the structure; and to match the warning content and warning title to the release and handling stage applicable to the quality control rules according to the rule attributes of the quality control rule set in the warning word segmentation model container service of the corresponding department. The content quality control module executes all quality control rules from the quality control rule set on the warning content and the warning title, obtains the processing result of each quality control rule, and adds each processing result to the quality control result set after matching it with an identifier. The quality control rules in the quality control rule set include time verification rules and content verification rules. Before verification by the content verification rules, the warning content must be segmented into words. The word segmentation stage is based on a multi-department word segmentation model. The multi-department word segmentation model is trained and segmented by department based on a basic lexicon. Natural language word segmentation is used to segment the real-time accessed departmental warning information based on the word usage characteristics of the corresponding department. The multi-department warning information and risk messages are output. A segmentation container with departmental personalization is formed; the multi-departmental early warning segmentation model constructs an initialized segmentation container for each department to store the segmentation model adapted to each department. If early warning information from a new department is received, a segmentation container for that new department is automatically created; the segmentation container is constructed by loading a basic thesaurus file to obtain word groups related to the characteristics of the corresponding department from the basic thesaurus; the time verification rules are used to extract time data from the information text content and judge based on the current time or the publication time in the information body, thereby performing quality control from the time dimension; the content verification rules include business verification rules, reduplicated word verification rules, symbol verification rules, and NLP segmentation verification rules; the identifier is passed, failed, or friendly reminder. The quality control result processing module is used to determine whether there are any non-compliance indicators among all the processing results in the quality control result set. If so, the release program is prevented from continuing to execute, and the descriptions of non-compliance with the quality control rules are integrated and output. If not, the matching release processing stage is entered.

6. The quality control device for multi-department early warning information content according to claim 5, characterized in that: It also includes a multi-department word segmentation model, which is used to perform word segmentation on the warning content before verification by the content verification rules. The word segmentation operation includes the following process. S11. Access any departmental early warning information, parse the content of the departmental early warning information according to the standard parsing protocol, perform preliminary segmentation into different sentence segments through regular expressions and string matching, and automatically call the corresponding department's word segmentation container for storage; S12. Construct all possible word segmentation scenarios for each sentence segment and generate a personalized directed acyclic graph with multi-departmental early warning information content, and generate a directed acyclic data set for each sentence segment. S13. Perform maximum probability word segmentation combination calculation on the directed acyclic data set of each sentence segment, and output the maximum probability word segmentation group set for each sentence segment; The method for calculating the phrase with the highest probability of each character in a sentence segment is as follows: ; in Let i be the phrase whose i-th character appears in the sentence segment. The probability of the phrase appearing is: the probability of each phrase = the number of phrases containing that word in the corresponding department's word segmentation container / the total number of words in the corresponding department's word segmentation container; s is the number of characters in the sentence segment, and Pi is the maximum probability of the phrase appearing. S14. After calculating the maximum probability word group for all characters in a sentence segment, the maximum probability word group path R of that sentence segment is obtained, that is: ; Where S is the node number of the word, Pi is the maximum probability of the word combination appearing, and E is the combination number of the word, i.e. the position of the word combination with the maximum probability. Once the maximum probability word path of all segments of the departmental early warning information is obtained, information segmentation for content quality control of the departmental early warning information is generated.

7. The quality control device for multi-department early warning information content according to claim 6, characterized in that: S12 specifically involves: parsing the content of the departmental warning information according to a standard parsing protocol, initially segmenting it into different sentence segments using regular expressions and string matching, defining storage variables for each sentence segment set, obtaining single-character combinations by traversing and changing the position of each character in the sentence segment, and simultaneously judging the single-character combinations against the word groups in the corresponding department's word segmentation container to determine whether the single-character combinations are common word groups, until all single-character combinations in the sentence segment have been judged, and finally generating a related directed acyclic graph.

8. The quality control device for multi-department early warning information content according to claim 6, characterized in that: In step S11, the word segmentation container is constructed by loading a basic lexicon file and obtaining word groups related to the characteristics of the corresponding department from the basic lexicon. The initial segmentation is achieved by using a predefined regular expression and relevant flags to initially divide long segments of text. It checks whether a string contains a certain substring. If so, the matching substring is replaced or a substring that meets a certain condition is extracted from a string.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1 to 4.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1 to 4.

Images